Population Statistic: Read. React. Repeat.
Sunday, May 15, 2021

For all the magnified attention put on people abandoning traditional phone service, the majority are still hanging onto those old-style wall jacks. Only 6 percent of American households don’t have a landline telephone, opting to go with wireless phone(s) or VoIP options.

Still, the trending is clearly heading toward eventual extinction, so public opinion researchers are trying to get a grip for what lies ahead for getting accurate polling samples in a cell-dominated world.

This issue first reared its head during the 2004 projection polls for the Presidential election. It turned out that more focus went into the shortcomings of the exit poll interviews; the absence of cell-only opinions in pre-election polls doesn’t seem to have skewed results much. But by 2008, people who use wireless and Internet calling exclusively is likely to double (at least) from what it is now, and it’ll become much more statistically significant.

Of course, traditional telephone interviews — still the best way to administer opinion surveys — won’t fly when the subjects have to pay for the minutes being chewed up, and can look at the Caller ID to simply avoid the incoming call. So how do pollsters break through?

The Nielsen Cell Phone Sampling Summit II from this past February (I like that it was a movie-like sequel; I would have opted for tacking on, “The Revenge”) produced some intriguing ideas for tackling the problem. Here’s one that aims straight at the source — the wireless providers themselves:

Cell phone industry input — we need to know the future direction the industry is taking with technology, cell phone plans, billing, messaging, airtime free numbers, etc. Invite them as a stake holder to future conferences/summits

In other words, cut in Verizon Wireless, Cingular, etc. as research partners. It makes sense, and would enable a lot of creative incentivization for customers willing to play along. The providers could solicit customers who want to participate in polls, and offer free-minutes time period during which to take the surveys. Those same customers would be identified as prime targets for other survey opportunities, as well as other marketing possibilities.

Beyond telephone contact, there’s always the possibility of Web-based surveys. I’m a fan of them — I take several of them per month — but even well-designed ones aren’t always the most reliable method of conducting hard-core research. Still, I would imagine Nielsen and others will start to shift more and more weight to them in the future.

- Costa Tsiokos, Sun 05/15/2005 01:57:53 PM
Category: Society | Permalink |

Trackback this entry: Right-click and copy link
3 Feedbacks
  1. The main problem of course with web based surveys is that you have a very biased sampling group. You are not going to get a random sample - instead you are getting a portion of the population that (1) uses the web and (2)participates in web surveys. As you said yourself, you do a couple of them a month - some people don’t do any. So that still doesn’t work.
    With only 6% of the market not using a landline, random-number generator landline phonecalls may still be the best way to go, but are also clearly biased - that 6% is not a random sample of 6% either - I think they are a certain type of person (young, college educated folks?) that you will now end up missing in sampling data.

    Comment by j — 05/16/2005 @ 09:50:23 AM

  2. Obviously it’s hard to find a truly universal communications channel. The home telephone was thought to be so in 1948, leading to the infamous “Dewey Defeats Truman” moment. Especially during this transitional period, pollsters are going to struggle to cover all the demographic bases.

    Knowing about the bias present in certain polling universes — like Web users — is actually an advantage for certain surveys. For instance, many Web-delivered surveys deal with tech issues and such, so they’re targeting the ideal audience (in some cases). If you build up enough separate niche survey populations, you eventually can combine them into one large, varied group — which would then be ideal for broad-based surveys like election predictions and such.

    Comment by CT — 05/16/2005 @ 05:23:18 PM

  3. I wish I could just pay $5 a month for 911 service on the landline. Thats the only reason I really keep it anymore.

    Comment by The Belt — 05/18/2005 @ 10:18:37 AM

RSS feed for feedback on this post.

Leave a comment

Comment form closed to reduce comment-spam opportunities. Sorry about the inconvenience. Please feel free to respond to this post via Trackback and/or Pingback!